The Center will continue the development of NAMD, a parallel molecular dynamics code for high-performance simulation of large biomolecular systems. NAMD supports mainstream biomedical research into the dynamics of cellular processes at atomic and sub-nanosecond resolution not achievable by experimental methods, and is used typically hand-in-hand with cryo-electron microscopy imaging and other experimental techniques. The largest NAMD simulations currently encompass entire viruses and cellular organelles. NAMD supports classical molecular dynamics simulations, most commonly of all-atom models with explicit sol- vent and periodic boundary conditions with PME full electrostatics in an NPT ensemble, although coarse-grained models, implicit solvent, and non-periodic or semi-periodic boundary conditions are also supported. CHARMM and similar force ?elds are supported, including the Drude polarizability model. Built on this foundation are a variety of features for steering simulations, including the ability to de?ne collective variables as control parameters for restraints or conformational free energy calculations. Also supported are alchemical free energy calculations and methods for accelerating sampling, including user-customizable multiple-copy algorithms for parallel tempering and conformational or alchemical free energy calculations. NAMD employs the prioritized message-driven execution capabilities of the Charm++/Converse parallel run- time system enabling excellent parallel scaling on both Cray and IBM supercomputers and commodity workstation clusters. NAMD is distributed free of charge as both source code and convenient precompiled binaries for Windows, Mac OS X, and Linux, and supports both NVIDIA GPUs and Intel Xeon Phi coprocessors. The performance of NAMD simulations will be improved by optimizations for current and future generations of GPUs such as NVIDIA Pascal and Volta, as well as the increasingly high core counts and wide vector instructions of CPUs such as Intel Xeon and Xeon Phi. NAMD will be ported and tuned on pre-exascale supercomputers including the Xeon Phi-based Theta and Aurora at Argonne National Laboratory and the GPU-based Summit at Oak Ridge National Laboratory, enabling up to multi-billion atom simulations of entire viruses and organelles, and building towards simulations of entire minimal cells on exascale supercomputers planned for 2023. The accuracy of NAMD simulations will be improved by the implementation of explicit solvent constant pH ensemble simulation and by combined quantum-classical simulation via a new interface to the quantum chemistry programs ORCA and MOPAC. Statistical con?dence in simulation results will be improved by enhanced sampling methods, in particular advanced multi-copy algorithms including exchanges with coarse-grained representations, milestoning, generalized simulated annealing, and automated reaction coordinate search. The development of these and other methods will be simpli?ed by the integration of features from the biomolecular visualization and analysis program VMD and an expanded interface to the Python scripting language.